ICML2020 å ææšè«ç³»è«æ èè çºè¡šäŒ (ãªã³ã©ã€ã³)
ã€ãã³ãå 容
æŠèŠ
æ©æ¢°åŠç¿ã«é¢ãããããåœéäŒè° International Conference on Machine Learning (ICML2020)ã§ãå€ãã®å ææšè«ã«é¢é£ããè«æãæ¡æãããŸãããæ¬å匷äŒã§ã¯ããã®ãã¡æ¥æ¬äººã«ãã£ãŠå·çãããè«æã«é¢ããŠãèè æ¬äººããæ¥æ¬èªã§ãã£ã·ãæéããšã£ãŠè«æã®èæ¯ãå 容ãå ±æããŸãã
ãªããæ¡æãããè«æäžèЧã¯ãã¡ãã«ãããŸãã
æ©æ¢°åŠç¿ and/or å ææšè«ã«é¢ããè«æãæ®æ®µèªãã§ããç ç©¶è ãåŠçããšã³ãžãã¢ãããŒã¿ãµã€ãšã³ãã£ã¹ãã®æ¹ãæ³å®ããŠãããŸããã誰ã§ããåå ããã ããŸãã
宿œæ¹æ³
ZOOMã䜿çšããŸããåèªã€ã³ã¹ããŒã«ããé¡ãããŸãã URLã¯åœæ¥ïŒ7/22ïŒã«ãconnpassã®ã¡ãã»ãŒãžæ©èœã§ç»é²è ã«äºåã«ãéãããããŸãã äžã€ã®çºè¡šæéã¯çŽ40åã§ãçºè¡šè³æïŒæ¥æ¬èª or è±èªïŒã¯ã€ãã³ãåŸã«å ¬éããäºå®ã§ãã
質çå¿çã¯ãsli.doã䜿çšããŠè¡ããŸãã äœ¿çšæ¹æ³ã¯ä»¥äžã®éãã§ãã
- connpassã®ã¡ãã»ãŒãžæ©èœã§ãç¥ããããURLã«ã¢ã¯ã»ã¹ããŠãã ãã
- çºè¡šè ã«è³ªåãããã°ãéœåºŠããã«æžã蟌ãã§ãã ããïŒå¿åå¯ïŒã
- 質åã¯ä»ã®æ¹ãé²èЧå¯èœã§ããèªåãèããããšãã質åãããã°ããããããã§ããŸãã
- çºè¡šçµäºåŸæéãèš±ãéããçºè¡šè ãããããããå€ããã®ããåªå ããŠåçããŠãããŸãã
泚æäºé
æè¡äº€æµãç®çã®å匷äŒã§ãã®ã§ãç¥èã®å ±æããã³ãåå è å士ã®äº€æµãç®çãšããªãåå ã¯ãæãããŠããŸãã åå ç®çãäžé©åã ãšå€æãããå Žåã«ã¯ãéå¶åŽã§åå ããã£ã³ã»ã«ãããŠããã ãå ŽåãããããŸãã
ã¿ã€ã ããŒãã«
| æé | å 容 |
|---|---|
| 19:00 - 19:05 | æšæ¶ã»è«žé£çµ¡ |
| 19:05 - 19:45 | çºè¡šâ ïŒCounterfactual Cross-Validation |
| 19:45 - 20:25 | çºè¡šâ¡ïŒFew-shot Domain Adaptation by Causal Mechanism Transfer |
| 20:25 - 21:05 | çºè¡šâ¢ïŒStatistically Efficient Off-Policy Policy Gradients |
â» é©å®äŒæ©ãåããŸããåœæ¥äºåãªãæéé åãå 容ã倿Žã«ãªãå¯èœæ§ãããããŸãã
çºè¡šã®è©³çް
ã¿ã€ãã«ïŒCounterfactual Cross-Validation: Stable Model Selection Procedure for Causal Inference Models
çºè¡šè ïŒéœè€åªå€ª (æ±äº¬å·¥æ¥å€§åŠ çµå¶å·¥åŠç³» åŠå£«èª²çš4幎)
ãããã£ãŒã«ïŒ äž»ã«å ææšè«ã𿩿¢°åŠç¿ã®èåæè¡ãçšããæ å ±æ€çŽ¢ã·ã¹ãã ã®ãã€ã¢ã¹é€å»ã«é¢ããç ç©¶ãè¡ãããŸããCyberAgent, Sony, ZOZO, SMNãªã©ã®åœå äŒæ¥ãšé£æºããŠãå ææšè« x æ©æ¢°åŠç¿é åã®ç€ŸäŒå®è£ ãå®èšŒç ç©¶ãé²ããŠããã
è«æãªã³ã¯: https://arxiv.org/abs/1909.05299
ã¹ã©ã€ããªã³ã¯ïŒhttps://speakerdeck.com/usaito/counterfactual-cross-validation-stable-model-selection-procedure-for-causal-inference-models-gong-kai-yong
è«ææŠèŠïŒ æ©æ¢°åŠç¿çãªæ±å誀差æå°åã®å®åŒåã«åºã¥ãå æå¹æäºæž¬ææ³ãä¹±ç«åŸåã«ãããŸãããŸããããã®ææ³ã¯ãå€ãã®ãã€ããŒãã©ã¡ãŒã¿ãæããŠããŸããããããèæ¯ãããåç°å¢ã«ã€ããŠæé©ãªäºæž¬ææ³ãšãã€ããŒãã©ã¡ãŒã¿ãéžæãããšããæé ã®éèŠæ§ãå¢ããŠããäžæ¹ã§ããã®æé ã«é¢ããç ç©¶ã¯ããŸãé²ãã§ããŸãããæ¬ç ç©¶ã§ã¯ã芳枬å¯èœãªããŒã¿ã®ã¿ãçšããŠããŒã¿ããªãã³ã«ãå æå¹æäºæž¬ææ³ã®ã¢ãã«éžæããã€ããŒãã©ã¡ãŒã¿ãã¥ãŒãã³ã°ãè¡ãããã®æ¹æ³ãææ¡ããŸãããŸãããã³ãããŒã¯ããŒã¿ã»ãããçšããŠãæ¢åã®ãã¥ãŒãªã¹ãã£ãã¯ãªè©äŸ¡ææšãããææ¡ææ³ãããè¯ãæ§èœãæã€ã¢ãã«ãåè£éåã®äžããæ¢ãåœãŠãããããšãå®èšŒããŸãã
We study the model selection problem in conditional average treatment effect (CATE) prediction. Unlike previous works on this topic, we focus on preserving the rank order of the performance of candidate CATE predictors to enable accurate and stable model selection. To this end, we analyze the model performance ranking problem and formulate guidelines to obtain a better evaluation metric. We then propose a novel metric that can identify the ranking of the performance of CATE predictors with high confidence. Empirical evaluations demonstrate that our metric outperforms existing metrics in both model selection and hyperparameter tuning tasks.
ã¿ã€ãã«ïŒFew-shot Domain Adaptation by Causal Mechanism Transfer
çºè¡šè ïŒæå¶æ¯ å¿ (æ±äº¬å€§åŠå€§åŠé¢ æ°é å嵿ç§åŠç ç©¶ç§ å士課çš2幎)
ãããã£ãŒã«ïŒ å°æ°ããŒã¿ããçµ±èšçæ©æ¢°åŠç¿ãè¡ãæ¹æ³è«ã®çè«çç ç©¶ãè¡ã£ãŠããŸãïŒç¹ã«å æçæ å ±ãåŠç¿ã«æŽ»çšãããšããã¢ãããŒãã®ç ç©¶ãé²ããŠããŸãïŒ
è«æãªã³ã¯: https://arxiv.org/abs/2002.03497
ã¹ã©ã€ããªã³ã¯ïŒhttps://takeshi-teshima.github.io/talks/2020-07-22/few-shot-domain-adaptation-by-causal-mechanism-transfer.pdf
è«ææŠèŠïŒ æ¬ç ç©¶ã§ã¯æå ã®ããŒã¿ãå°æ°ãããªãå Žåã«ã粟床ã®è¯ãäºæž¬åšãåŠç¿ããæ¹æ³è«ãšããŠããã¡ã€ã³é©å¿ (Domain adaptation; DA)ãã®å®çŸæ¹æ³ãèããŸãïŒãã¡ã€ã³é©å¿ãšã¯ïŒæå ã®ããŒã¿ãéãããŠãããšãã«ãæå ã®ããŒã¿ãšã¯ç¢ºçååžãç°ãªããé¢é£æ§ã®ãã远å çããŒã¿ããåŠç¿ã«åœ¹ç«ãŠããšããæ¹æ³è«ã§ãïŒãã¡ã€ã³é©å¿ã®æ¹æ³ãéçºããäžã§æãéèŠãªåãã¯ïŒããé¢é£æ§ããšããŠç°ãªãããŒã¿éã«ã©ã®ãããªé¢ä¿ãä»®å®ãããïŒè»¢ç§»ä»®å®ïŒãã§ãïŒæ¬ç ç©¶ã§ã¯è»¢ç§»ä»®å®ãšããŠãããŒã¿ååžã®èåŸã«ãããå æã¢ãã«ããå ±éã§ããããšããä»®å®ãçšããããšãã§ããå¯èœæ§ãæ¢çŽ¢ããŸããïŒäŸãã°ä»®æ³çå¿çšäŸãšããŠå»çèšé²ããçŸç äºæž¬ãè¡ãããã®äºæž¬åšãïŒããå°åã«ç¹åããŠåŠç¿ããããšããç¶æ³ãèããŸãïŒãã®éïŒç°ãªãå°åã§ã®ããŒã¿ã掻çšããããšããŠãïŒçæŽ»ç¿æ £ãç°ãªããªã©ã®çç±ã§ããŒã¿ã®ååžãã®ãã®ã¯å€§ããç°ãªãå¯èœæ§ããããŸãïŒããããªããåãçŸç ã¯å°åã«ãããåãæ©åºãæã€ãšæåŸ ããããšã¯ã§ããŸãïŒãããã£ãç¶æ³äžã§ãèåŸã«ããå æçæ©æ§ãåäžããšããäºåç¥èãå©çšããŠç²ŸåºŠã®è¯ãåŠç¿ãè¡ãæ¹æ³ãæäŸããããšãæ¬ç ç©¶ã®çæ³çãŽãŒã«ã§ãïŒæ¬è«æã§ã¯ïŒå ·äœçã«ã¯åãã¡ã€ã³ã®ç¢ºçååžã®èåŸã«æ§é çå æçã¢ãã«ããããšä»®å®ããããã§ïŒããã®æ§é æ¹çšåŒãå ±éã§ããããšãã転移仮å®ãçšããããšãææ¡ããŸããïŒãã®ä»®å®ãå©çšããŠãã¡ã€ã³é©å¿ãè¡ãæ¹æ³ãéçºãïŒãã®çè«çè§£æãéããŠææ¡æ³ãçµ±èšçåŠç¿ã«ãããŠã©ã®ããã«è²¢ç®ããããæããã«ãïŒããã«Proof-of-conceptãšãªãå®ããŒã¿å®éšãéããŠææ³ã®åŠ¥åœæ§ã確èªããçµæãå ±åããŠããŸãïŒ
We study few-shot supervised domain adaptation (DA) for regression problems, where only a few labeled target domain data and many labeled source domain data are available. Many of the current DA methods base their transfer assumptions on either parametrized distribution shift or apparent distribution similarities, e.g., identical conditionals or small distributional discrepancies. However, these assumptions may preclude the possibility of adaptation from intricately shifted and apparently very different distributions. To overcome this problem, we propose mechanism transfer, a meta-distributional scenario in which a data generating mechanism is invariant among domains. This transfer assumption can accommodate nonparametric shifts resulting in apparently different distributions while providing a solid statistical basis for DA. We take the structural equations in causal modeling as an example and propose a novel DA method, which is shown to be useful both theoretically and experimentally. Our method can be seen as the first attempt to fully leverage the structural causal models for DA.
ã¿ã€ãã«ïŒ Statistically Efficient Off-Policy Policy Gradients
çºè¡šè ïŒäžåé ä¿ (Harvard倧åŠçµ±èšåŠç§ å士課çšïŒå¹Ž)
ãããã£ãŒã«ïŒ 匷ååŠç¿ãšå ææšè«ã®å¢çãäž»ã«ç ç©¶ããŠããŸãã
è«æãªã³ã¯: 1. https://arxiv.org/abs/2002.04014 2. https://arxiv.org/abs/1908.08526
è«ææŠèŠïŒ å»åŠãçµæžåŠã®å¿çšåéã§ãéå»ã®æç³»åããŒã¿ãçšããŠãæ¹çãè©äŸ¡ãæé©ãªæ¹çãåŠç¿ããææ³ïŒOff-policy evaluation, Off-policy learningïŒã¯éèŠã«ãªã£ãŠããŠããŠããããŸãçè«çãªç ç©¶ãå ææšè«ã匷ååŠç¿ã®ã³ãã¥ããã£ã§ãšãŠãçãã«ãªã£ãŠãããããããæ¢åã®æåãªæšå®ææ³ïŒSequential IPWã Marginal sturctural modelïŒã¯ãã©ã€ãºã³æ°ãé·ããªãã«ã€ãæšå®éã®èª€å·®ãææ°çã«ççºãããšããããã©ã€ãºã³æ°ã®åªããšããåé¡ããããå®éã«ãã¢ãã€ã«ã¢ããªãçšããMicro randomized trialsã§ã¯ãã©ã€ãºã³ãäœçŸãšãªãããã©ã€ãºã³æ°ã®åªãã¯æ·±å»ãªåé¡ã«ãªãã æ¬ç ç©¶ã§ã¯MDPã«ãããæ¹çè©äŸ¡ãšæ¹çåŸé è©äŸ¡ã®èª€å·®ã®æŒžè¿äžéãå°åºããå®ã¯ãã®äžéã¯ãã©ã€ãºã³æ°ã«ææ°çã§ã¯ãªãå€é åŒçã«äŸåããããšã瀺ããããããŠããã®äžéãéæããæšå®éãææ¡ããã©ã€ãºã³æ°ã®åªããïŒéšåçã«ïŒè§£ãããææ¡ææ³ã¯äºã€ã®æšå®éïŒåšèŸºå¯åºŠæ¯ãšQ颿°ïŒãçµã¿åãããã¡ã¿ã¢ã«ãŽãªãºã ã§ãããDoublyRobustãšããããã¹ãæ§ãæã€ããããŠææ¡ããæ¹çåŸé æšå®ãçšããåŠç¿ææ³ã®åæã¬ãŒãã瀺ãããã¯ãã¬ãŒãããã©ã€ãºã³æ°ã«å¯ŸããŠå€é åŒçã«äŸåããããã«ã§ããããšã瀺ããããŸãæ¬çºè¡šã§ã¯è¿å¹Žã®å ææšè«ãšåŒ·ååŠç¿çéã«ãããOff policy learningã®éèŠãªè«æãã¡ãšã®é¢é£ããæéãèš±ãéãã俯ç°çã«çºè¡šããäºå®ã§ããã
1. Policy gradient methods in reinforcement learning update policy parameters by taking steps in the direction of an estimated gradient of policy value. In this paper, we consider the statistically efficient estimation of policy gradients from off-policy data, where the estimation is particularly non-trivial. We derive the asymptotic lower bound on the feasible mean-squared error in both Markov and non-Markov decision processes and show that existing estimators fail to achieve it in general settings. We propose a meta-algorithm that achieves the lower bound without any parametric assumptions and exhibits a unique 3-way double robustness property. We discuss how to estimate nuisances that the algorithm relies on. Finally, we establish guarantees on the rate at which we approach a stationary point when we take steps in the direction of our new estimated policy gradient.
2. Off-policy evaluation (OPE) in reinforcement learning allows one to evaluate novel decision policies without needing to conduct exploration, which is often costly or otherwise infeasible. We consider for the first time the semiparametric efficiency limits of OPE in Markov decision processes (MDPs), where actions, rewards, and states are memoryless. We show existing OPE estimators may fail to be efficient in this setting. We develop a new estimator based on cross-fold estimation of q-functions and marginalized density ratios, which we term double reinforcement learning (DRL). We show that DRL is efficient when both components are estimated at fourth-root rates and is also doubly robust when only one component is consistent. We investigate these properties empirically and demonstrate the performance benefits due to harnessing memorylessness.
泚æäºé
â» ãã¡ãã®ã€ãã³ãæ å ±ã¯ãå€éšãµã€ãããååŸããæ å ±ãæ²èŒããŠããŸãã
â» æ²èŒã¿ã€ãã³ã°ãæŽæ°é »åºŠã«ãã£ãŠã¯ãæ å ±æäŸå ããŒãžã®å 容ãšå·®ç°ãçºçããŸãã®ã§äºããäºæ¿ãã ããã
â» ææ°æ å ±ã®ç¢ºèªãåå ç³èŸŒæç¶ããã€ãã³ãã«é¢ãããåãåããçã¯æ å ±æäŸå ããŒãžã«ãŠãé¡ãããŸãã

ãåãåãã
é¢é£ããã€ãã³ã

ãŒãããã¯ãããPythonå ¥éè¬åº§ïŒå¯Ÿé¢ã»ç¡æïŒ
2026/11/07(å) éå¬
ãŒãããã¯ãããPythonå ¥éè¬åº§ïŒå¯Ÿé¢ã»ç¡æïŒ
2026/11/28(å) éå¬
AIãšã³ãžãã¢ã®ä»äºãäœéšïœClaude Codeã§ã売äžããŒã¿ãè§£æã»äºæž¬ã§ããAIã¢ããªãäžç·ã«äœã
2026/11/05(æš) éå¬
DataScience&DataEngineering MeetUp #3
2026/11/24(ç«) éå¬
ããªã³ããã³ãé ä¿¡ãAI/DS Case Talk#02ããŒã¿ãµã€ãšã³ãã£ã¹ãã«æ±ãããããããŒã¿ã®è£ãèªã¿ã¬ãåããšã¯ïŒ
2026/09/28(æ) éå¬- TOP
- ã€ãã³ã
- ICML2020 å ææšè«ç³»è«æ èè çºè¡šäŒ (ãªã³ã©ã€ã³)
